FAIRBORN — Researchers at Wright State were right: They accurately predicted that Donald Trump would best Hillary Rodham Clinton in the general election to become the 45th president of the United States.
On the night of the third presidential debate (Oct. 19 in Las Vegas), members of the Applied Policy Research Institute tracked real-time Twitter sentiment about the businessman and former secretary of state using “TWITRIS,” which analyzes tweets send to Twitter.
When researchers used it again on Election Day, the feelings of tweeters early on pointed to the Republican candidate.
“What we saw during Election Day is Trump’s sentiment was much more steady and consistent than Hillary’s, especially in battleground states,” institute director Mike Wiehe told News Center 7’s Natalie Jovonovich on Thursday afternoon. Clinton’s “was much more volatile, which would be indicative potentially of a non-motivated voter base.”
The APRI at WSU studies social sciences, and that includes trends in politics. WSU researchers monitored all three debates and for that third debate hosted a campus event in collaboration with Kno.e.sis and local start-up Cognovi Labs.
Amit Sheth, founder of Cognovi Labs, went so far as to say “polling is dead, asking people what would you do doesn’t work. Listening to what they volunteer works.”
Sheth said, “Social media is an active tool, you have to participate in it to capture the data. You get even better information without doing the exit poll and in real time, at any time.”
APRI got the 2016 general election right because “we were tracking what people were actually saying that were actively doing something online and we could see it,” Sheth said. “And in those battleground states, you could just see that those lines were so much more volatile for Hillary Clinton compared to Trump that she wasn’t going to carry those states.”
Cognovi Labs used “TWITRIS” to successfully call the Brexit vote as well (the United Kingdom’s withdrawal from the European Union in June).
APRI’s plan is to collect all of their data and present it to elected officials to show them how to effectively poll people for 2020.